From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nix Subject: Re: [SCSI REGRESSION] 3.10.2 or 3.10.3: arcmsr failure at bootup / early userspace transition Date: Tue, 30 Jul 2013 01:56:20 +0100 Message-ID: <87ob9koogb.fsf@spindle.srvr.nix> References: <87r4ehfzhf.fsf@spindle.srvr.nix> <51F667C2.4020801@fastmail.fm> <87mwp5frdl.fsf@spindle.srvr.nix> <51F67959.2060803@fastmail.fm> <87fvuxdqes.fsf@spindle.srvr.nix> <51F708A4.9090207@interlog.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: In-Reply-To: <51F708A4.9090207@interlog.com> (Douglas Gilbert's message of "Mon, 29 Jul 2013 20:28:20 -0400") Sender: stable-owner@vger.kernel.org To: Douglas Gilbert Cc: Bernd Schubert , Linux Kernel Mailing List , linux-scsi@vger.kernel.org, "Martin K. Petersen" , nick.cheng@areca.com.tw, stable@vger.kernel.org List-Id: linux-scsi@vger.kernel.org On 30 Jul 2013, Douglas Gilbert outgrape: > Please supply the information that Martin Petersen asked > for. Did it in private IRC (the advantage of working for the same division of the same company!) I didn't realise the original fix was actually implemented to allow Bernd, with a different Areca controller, to boot... obviously, in that situation, reversion is wrong, since that would just replace one won't- boot situation with another. It looks like a solution is possible that will let us boot *both* my controller (with its old 2009-era firmware) *and* his. We just have to let Martin implement it. Give him time, I only got a successful boot out of it an hour ago :) > I just examined a more recent Areca SAS RAID controller > and would describe it as the SCSI device from hell. One solution > to this problem is to modify the arcmsr driver so it returns > a more consistent set of lies to the management SCSI commands that > Martin is asking about. I can't help notice that something is skewy in its error handling, too. When the controller errors, even resetting the bus doesn't seem to be enough to bring it back :/ I've seen errors from it before which did *not* lead to it imploding forever, but this is apparently not one such. Certainly Areca-the-company has... issues with communication with the community (i.e., they don't). A shame I didn't know that before I bought the controller and made all my data completely dependent on it, really. Shame, the controller otherwise works very well (fast, and has coped with a disk failure with aplomb). -- NULL && (void)